That’s the first high quality diagram I’ve seen directly in the thread, not at some off-board paywalled cite/site.
Thank you!
There’s room for that diagram to be significantly simplified, but it’s interesting that as depicted there’s one turquoise-colored lock bolt at the door bottom but the photos we’ve seen upthread of the UAL door with loose hardware there were two adjacent loose bolts in the housing just above the big spring at the bottom.
Or maybe the loose bolts we’ve seen aren’t themselves the locking bolts but are yet other bolts holding par of the overall dummy plug assembly together. Gets more interesting with each bit of evidence we receive.
The loose bolts seen in pictures were the ones mounting the lower spring bolt assembly to the door. .I believe there are 4 each for the 2 bottom assemblies. What I haven’t seen is what secures them in place. Are they torqued down or are there crown nuts or drilled nuts so they can be locked in place.
You can also watch the video below (Mentour Now…he is a 737 pilot) for a good discussion and detail on how those doors are supposed to work. Jump to 10 minutes in where he starts getting more into the details of how those doors work.
He does note, towards the end of the video, that this door design has long been used with no issue so it probably is not a design flaw but, rather, a manufacturing or quality control problem.
737-800 forced to return to airport Sunday in Japan after cracks develop in cockpit windshield. Normally this wouldn’t be a big story but it just adds to the “Boeing can’t catch a break” story.
One of Boeing’s PR problems, rather ironically, is that the 737, across all models, is one of the most popular airliners ever built, with nearly 12,000 delivered and thousands more on order, and with that many flying something is bound to go wrong with one of them somewhere sometime. I also remember reading that, at least prior to the two crashes that grounded the entire MAX fleet worldwide, the 737 also had one of the best safety records ever. But this may serve only to draw a distinction between the old Boeing – the one that was regarded as the “quality” alternative to McDonnell-Douglas – and the new trouble-plagued one.
Windshield failures aren’t exactly common, but they are not rare either. I personally never had one but I know several pilots who have.
The design has multiple layers and the one that provides the strength is usually not the one that cracks. Typically it’s the outer layer that’s exposed to extreme thermal cycling and the occasional bird impact plus long-term sustained shot-peening by rain that cracks. Seeing through it is really the only hard part.
They tell us there’s not really much incremental hazard associated with flying around with a cracked outer windshield.
Whether you continue to the destination or land early or return to where you took off is mostly a matter of personal preference. Significant weather or airfield differences at either end might tilt the scale too.
On extremely rare occasion the inner (strength) layer cracks. That is a bigger deal and now it’s time to descend to a safe altitude and depressurize. Which may in turn drive your decision on where to land based on reduced range.
All else equal, having a forward-facing windshield fail is a bigger deal than having a side window fail although they’re constructed in the same manner.
The details in the above are 737-specific, but the broad conceptual outline would apply to any airliner by any manufacturer.
I have some knowledge on crj wibdows (though not much, it’s all second hand) and while it’s true that the outer pane is more likely to fail, the design is such that each pane is certified to handle the expected flight loads on their own, which is pretty cool. From what I recall of a talk I attended about the cabin windows in particular is that the little hole in the one pane is there because the other pane is the primary one; it’s doing all the work unless it breaks, and then the loads are transferred to the other one.
There’s a fun little certification reality in that on older planes the requirements to have a clear view under precipitation conditions resulted in the pilot side window never being allowed to have a distorted view (no mmel relief), while the copilot side one did not have that restriction. It’s a weird little fallout of how particular designs were shown to comply and the way older versions of the standards were written. I’ve seen application letters to regulatory authorities to allow flight with small side window scratches, provided the copilot is flying during TT&L. It’s almost always accepted to my knowledge.
In the video he noted that thankfully the window and middle seats next to the door plug were unoccupied. Generally when I’m assigned an aisle seat and then discover I’ve scored an entire group of three seats to myself I slide over to the window seat before we leave gate. I’m sure the person sitting in the aisle seat in that flight is glad they didn’t do that.
Not really. The 717 is branded as a Boeing, but it’s really a McDonnell-Douglass design. You could probably even say it’s a Douglass design, since the basic design goes back to the DC-9.
But even if they were both Boeing designs, how is that relevant here?
It was not immediately clear what caused the incident on Sunday and whether it was related to a manufacturing flaw.
I guess it’s plausible that there could have been some problem with one of the plane’s steering systems, but I am guessing that’s pretty unlikely, given that neither plane is particularly new. And I hate how the media always attempts to tie every minor incident involving a Boeing plane back the the previous more serious incident, no matter how tenuous the connection.
What we see starting here is a media trend. They watch the never-ending flood of minor shit that happens, and if it’s a Boeing it’s newsworthy, so play up the “Boeings are involved in yet another problem” angle.
Which could really get the ball rolling on the public choosing to believe they are not a safe product. Not the new ones, but also not the old ones. Which both of these aircraft are.
A taxi collision is typically about 95% pilots and 5% ATC. In this case it smells like All Nippon goofed and whacked Delta, but we don’t yet know if Delta was in an unexpected non-standard place when it happened. You can’t go willy-nilly driving around assuming everybody else is on their marks, but many airports, including O’Hare are really jam-packed and if you can’t see his marks, and he looks about in the right place, stick to your own marks and it should work. Should.
I wonder what the weather and ramp and taxiway conditions were like at O’Hare on Sun evening at 6:30. it was certainly a very busy time of day.
I had wondered if anyone would mention that famous event. That was a very different sort of failure. The windshield pane(s) didn’t fail; the whole assembly, frame and all, wasn’t properly attached to the airplane and departed as a unit. So rather like the recent MAX 9 dummy door plug; a whole assembly improperly attached to its surrounding structure.
A bad failure, but fundamentally a different one from stressed glass shattering.
It has always been quite amazing to me that in the BA event both the pilot and the airplane survived. I’ve often thought that had he gone all the way out, which was the way to bet, the wind in the cockpit would have incapacitated to FO too. Having the Captain as a (partial) plug in the opening saved the day. Or so I speculate.
Imagine if need reporters actually started reviewing all public documents such as airworthiness directives (there’s like, one more per day from the FAA alone) or the Transport Canada CADORS. They could really make the public freak out!
Good engineering does assume that failures will happen; it just also correctly identifies those failures and mitigates them.
They could, if they so chose, beat that drum hard enough to (probably) trigger a self-sustaining scare. Which would not be good for anyone except maybe media companies seeking clicks and eyeballs.
I wonder if a type of smart tool can be used to prevent this sort of thing.
Maybe it already exists, or is just impractical to make or use.
For instance. A job requires X number of fasteners to be torqued to Y . When starting the job the tool is initialized to that job. Either the tool has an indicator or it is linked via wireless, to report that it has done X torque cycles. Until that number of cycles is complete either the tool or an external system does not allow the job to be indicated as done. If more cycles are run, the job must get extra inspection as well.
I have worked with systems that have dedicated part trays. At finish, all the parts must be gone or the trays filled properly. They were not smart trays, but maybe such things can also be connected too. I assume most or all aircraft manufacturers at least use the custom tray system.
You can engineer a great safe thing. But if assembled badly it is negated.
Oh, those have been around for decades. Used 'em when I was a little one working the assembly line. Pretty simple, just an electric torque wrench and a computer that reads the torque. Pretty easy to cheat, though. If you don’t want to move the wrench to another bolt, just torque one of them over and over.
They’re useless without proper quality control, which is often one of the first place to make cuts for profit. It’s not like QC/QA actually MAKES anything.
You’d have to have a digital twin of the product, I think, to make that work? So accurate 3D models with exact hardware counts, for example. The baseline 737 is very old, entire sections of that aircraft likely have no 3D representation and everything depends on tooling to assemble correctly. Newer parts and components are likely done in CAD, but against dummy interface surfaces that aren’t the true parts.
I’ve designed repair and replacement parts for old aircraft (not Boeings) and needed to simulate contact surfaces and be very generous in the design to account for on-the-spot trimming, cold working angles, fastener pitch (“pick up existing holes” where I didn’t know for sure how many there were when installed 25 years ago, just that they’re at a standard pitch).
Newer aircraft have much more technology built into their entire design, manufacturing, quality control and maintenance processes, but old ones…I always joked they were built first and drawn after. There’s only so much you can do to modernize it when the parts being made are still based on hand-drawn mylar from 1972.
In this case, the door/plug structure was new(ish) so likely some improvement could be possible, assuming you could meld it with the legacy bits of your final assembly line.
There should be maintenance crews at every city the planes are at. How hard is it to go through a checklist and document if any of the doors were assembled incorrectly?
The problem is that the doors are a symptom. Unless they find a specific design defect in the door assembly, the problem is almost certainly sloppiness, poor quality control, problems in the supply chain with out of spec parts, or something like that. For example, if a bolt sheared that shouldn’t have there needs be be a thorough review of the paper trail to find out why, and if they discover a manufacturing flaw, then any bolts from the affected batches would have to be replaced. If they are common, they could be in many places.
The place to start is probably what they are doing - a review of all the QC procedures, handoffs, inspection records, paper trails for all affected parts, etc. Then based on what they find, go from there.
I’ll bet you they find a lot of issues in their review. Procedures not followed, issues signed off on when they shouldn’t have been, etc. There has been a lot of management pressure on engineering to constantly shave costs.
There are overnight maintenance crews at about 5% of each airline’s cities. The inspection process, once written and approved, is probably a 10- or 15-man hour job taking two people about 6 elapsed hours. First you remove the seats. Then you remove the interior paneling. Then …
Of course that takes care of the left door. Then you have to do it again on the right door.
As @Sam_Stone said, the issue is not that some particular nuts weren’t torqued. The issue is we now have little reason to believe that any nut on any jet, much less every nut on every jet, was properly torqued. Or at least we have little reason to believe that the documentation claiming each nut is properly torqued was properly sourced and is trustworthy.
Slovenly workmanship was allowed to go on. OR, there’s an inherent inadequacy in how those nuts were retained after proper installation. Bad engineering in a single instance is actually easier to recover from than bad assembly in an unknown number of instances.
